Quantifying Quality: A Computational Approach to Literary Value in North Korea

Abstract

This article analyzes questions of literary production and literary value in the Democratic People's Republic of Korea (DPRK) using quantitative methods and data sets covering a significant part of the country's literary and intellectual output over forty years (1977–2016). The first part of the article adopts a thematic approach, using topic modeling and time series, to map the DPRK's literary output and investigate the influence of genre on literary consecration. The second part looks at questions of style and literary quality in the DPRK through a comparative analysis of multiple novels grouped by degree of literary prestige. It introduces a method to evaluate the originality of an author's writing style using a masked language model (BERT) and uses it to assess the role of conformity and originality in the DPRK's literary field. Even though writers consecrated by the state display a high level of stylistic conformity when writing mandated biographies of the ruling family, their independent output ranks higher in originality than nonconsecrated writers.